Foliar respiration acclimation to temperature and temperature variable Q10 alter ecosystem carbon balance
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Global Change Biology (2005) 11, 435–449 doi: 10.1111/j.1365-2486.2005.00922.x Foliar respiration acclimation to temperature and temperature variable Q10 alter ecosystem carbon balance K I R K R . W Y T H E R S *, P E T E R B . R E I C H *, M A R K G . T J O E L K E R w and P A U L B . B O L S T A D * *Department of Forest Resources, University of Minnesota, 115 Green Hall, 1530 Cleveland Ave N., Saint Paul, MN 55108, USA, wDepartment of Forest Science, Texas A&M University, College Station, TX 77843-2135, USA Abstract The response of respiration to temperature in plants can be considered at both short- and long-term temporal scales. Short-term temperature responses are not well described by a constant Q10 of respiration, and longer-term responses often include acclimation. Despite this, many carbon balance models use a static Q10 of respiration to describe the short-term temperature response and ignore temperature acclimation. We replaced static respiration parameters in the ecosystem model photosynthesis and evapo-transpiration (PnET) with a temperature-driven basal respiration algorithm (Rdacclim) that accounts for temperature acclimation, and a temperature-variable Q10 algorithm (Q10var ). We ran PnET with the new algorithms individually and in combination for 5 years across a range of sites and vegetation types in order to examine the new algorithms’ effects on modeled rates of mass- and area-based foliar dark respiration, above ground net primary production (ANPP), and foliar respiration– photosynthesis ratios. The Rdacclim algorithm adjusted dark respiration downwards at temperatures above 18 1C, and adjusted rates up at temperatures below 5 1C. The Q10var algorithm adjusted dark respiration down at temperatures below 15 1C. Using both algorithms simulta- neously resulted in decreases in predicted annual foliar respiration that ranged from 31% at a tall-grass prairie site to 41% at a boreal coniferous site. The use of the Rdacclim and Q10var algorithms resulted in increases in predicted ANPP ranging from 18% at the tall- grass prairie site to 38% at a warm temperate hardwood forest site. The new foliar respiration algorithms resulted in substantial and variable effects on PnETs predicted estimates of C exchange and production in plants and ecosystems. Current models that use static parameters may over-predict respiration and subsequently under-predict and/or inappropriately allocate productivity estimates. Incorporating acclimation of basal respiration and temperature-sensitive Q10 have the potential to enhance the application of ecosystem models across broad spatial scales, or in climate change scenarios, where large temperature ranges may cause static respiration parameters to yield misleading results. Keywords: acclimation, ANPP ecosystem model, PnET, production, Rd : A, respiration, temperature Received 29 September 2003; revised version received 16 September 2004; accepted 21 October 2004 biosphere and the atmosphere. Carbon dioxide (CO2) Introduction flux is of particular concern because it is a greenhouse Concern about climate change (Houghton et al., 1992; gas and is the form in which most C moves between the Vitousek, 1994) and associated long-term impacts on biosphere and the atmosphere. Over the last 200 years, the planet (e.g. Lubchenco et al., 1991; Woodwell & atmospheric CO2 concentration has increased by 30% Mackenzie, 1995; Falkowski et al., 2000) has intensified (Keeling et al., 1996). A likely outcome of this change is interest in the flux of carbon (C) between the terrestrial an alteration in global temperature patterns. Estimated increases of 1 1.4 to 1 5.8 1C in mean global surface Correspondence: Kirk R. Wythers, fax 1 612 625 5212, temperature have been predicted to occur from 1990 to e-mail: kwythers@umn.edu 2100 (Houghton et al., 2001; McCarthy et al., 2001), and r 2005 Blackwell Publishing Ltd 435
436 K . R . W Y T H E R S et al. these changes are likely to vary substantially among a ratio between a respiration rate at one temperature regions and exhibit seasonal and diurnal variation. and the respiration rate at a temperature 10 1C lower Regardless of scale, atmospheric C is fixed into plant (Lavigne & Ryan, 1997; Bolstad et al., 1999; Tjoelker biomass through photosynthesis and returned to the et al., 2001; Atkin & Tjoelker, 2003; Larcher, 2003). atmosphere via respiration. The difference between Many carbon balance models use approaches that fix these fluxes determines C balance in ecosystems. While the Q10 at or near 2.0, and fix Rdref as a proportion of C is released back to the atmosphere through both photosynthesis (e.g. Ryan, 1991; Aber & Federer, 1992; autotrophic and heterotrophic pathways, autotrophic Melillo et al., 1993; Aber et al., 1995, 1996, 1997; Schimel respiration accounts for roughly half of the total et al., 1997; Cramer et al., 1999; Kimball et al., 2000; respiratory C flux (Farrar, 1985; Houghton et al., 2001). Sands et al., 2000; Stockfors, 2000; Clark et al., 2001; Therefore, autotrophic respiration plays a substantial Potter et al., 2001a,b; Sampson et al., 2001; Sitch et al., role in governing ecosystem C balance (Field et al., 1992; 2003). However, respiration response to short- and Ryan et al., 1995, 1996). Accurate modeling of the long-term changes in temperature often do not follow a response of autotrophic respiration to changing climate simple exponential Q10 (Gifford, 2003; Larcher, 2003). In will be essential in order to effectively predict the particular, two issues which may be problematic under impact of climate change on the global C balance. Given such approaches are: (1) the degree to which short term the established links between plant respiration and (seconds to minutes) dark respiration response to temperature, it may be useful to re-examine the temperature departs from a simple exponential func- mechanisms and assumptions built into process or- tion (Wager 1941, Tjoelker et al., 2001) and (2) the fact iented ecosystem models, common tools for evaluating that acclimation to temperature may shift the entire the effects of climate on C balance patterns over a temperature response function (regardless of its shape) variety of temporal and spatial scales. (Atkin & Tjoelker, 2003). C balance models that operate at the tissue level of Although respiration responds to temperature on organization (i.e. leaf, stem, and root) are often used to both short- and long-term time scales, the Q10 of examine feedbacks between environmental change and respiration describes the short-term sensitivity of ecosystem productivity. Models that simulate system respiration to temperature (i.e. seconds to hours). The behavior in terms of a C balance typically do so with a near-instantaneous exponential respiration function collection of interactive algorithms that estimate C described by Q10 has been shown to inadequately fit assimilation, respiration, and allocation. For a more empirical observations in plants (Belehrádek, 1930; complete review of process models see Ågren et al. Wager, 1941) (reviews by James, 1953; Forward, 1960; (1991), Ryan et al. (1996), and Mäkelä (2000). While most Berry & Raison, 1981) and soils (Lloyd & Taylor, 1994). process models incorporate temperature into their Nevertheless, no general alternative had been pro- respiratory calculations, they do so at varying levels posed. Recently, however, Tjoelker et al. (2001) showed of complexity, using a range of assumptions and that the observed responses could be better fit with a generalizations. The difficulty in capturing plant quasi-exponential function whose exponent varied with respiration’s many sources of variation in a mathema- temperature. Tjoelker et al. (2001) synthesized the tical model owes to the debate in how to best represent results of published foliar Q10 of respiration values (or to represent at all) these sources of variation. A across a range of plant taxa (grasses, forbs, and woody better understanding of how plant respiration responds plants) and across a range of biomes (tropical, to temperature change can only aid in the construction temperate, boreal, and arctic), and concluded that the of more flexible and generalizable algorithms. respiratory Q10 declined linearly with increasing Many biological processes, including respiration, are measurement temperatures in a consistent manner dependent upon temperature. Biological processes are among a range of taxa and climactic conditions. In often ascribed to Van’t Hoff’s reaction rate-temperature essence, dark respiration exhibits decreasing Q10 values rule, and modeled as exponential functions. The (measured over 5–10 1C intervals) with increasing respiration–temperature response function is such an measurement temperature and the response appears exponential relationship. A form of the respiration consistent among species, and across diverse biomes response function which permits direct estimates of the (Tjoelker et al., 2001). This evidence suggests that the Q10 parameter is often expressed as: use of a static Q10 of 2.0 is inappropriate for large temperature ranges. ½ðTTrefÞ =10 In addition to near-instantaneous temperature re- Rd ¼ Rdref Q10 ; ð1Þ sponse, rates of respiration are known to acclimate to where Rd is dark respiration, Rdref is the specific thermal environment over longer time periods (days to respiration at a reference temperature ( 1C), and Q10 is months) through adjustments in the overall elevation of r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
F O L I A R R E S P I R AT I O N A C C L I M AT I O N A LT E R S E C O S Y S T E M C A R B O N B A L A N C E 437 the temperature response function. Respiratory C Table 1 Process-based ecosystem models and their primary exchange rates are known to acclimate with time to literature sources prevailing temperatures in plant leaves (Larigauderie & Model Principal literature Körner, 1995; Tjoelker et al., 1999a, b; Atkin et al., 2000a), roots (Gunn & Farrar, 1999; Tjoelker et al., 1999a), soils 3-PGz Landsberg & Waring (1997) (Luo et al., 2001), and ecosystems (Enquist et al., 2003). BEPSw Liu et al. (1997) Such acclimation can be substantial (Gunderson et al., BIOMASSw McMurtrie et al. (1990) 2000) and rapid (Atkin et al., 2000a, b; Bolstad et al., BGC familyw Running & Coughlan (1988) 2003), and therefore have significant effect on C CENTURYw Parton et al. (1987) balance. Acclimation to temperature may result from COCA/FEFw Hari et al. (1999) a change in Q10, a shift in the elevation (intercept) of the COMMIX Bartelink (2000) FINNFORw Këllomaki & Väisänen (1997) temperature response function, or a combination of FORDYNw Luan et al. (1996) both (Tjoelker et al., 1999b; Atkin et al., 2000a, b; Atkin & FORGROw Mohren & Kramer (1997) Tjoelker, 2003). GOTILWAw Gracia et al. (1999) Typically, temperature acclimation to a warmer LPJw Sitch et al. (2003) environment results in a downward shift of the short- NASA-CASA Potter et al. (2001a, b) term temperature response function. This shift is PnET familyw Aber et al. (1996) reflected in decreased respiration at a standard tem- PROMODw Sands et al. (2000) perature in warm acclimated plants compared with SECRETSw Sampson et al. (2001) cold acclimated plants. Consequently, the response of SPAMw Frolking et al. (1996) respiration to variation in thermal environment (over SPAM2w Clark et al. (2001) days to seasonal) will differ from predictions based on TEM Raich et al. (1991) short-term temperature response functions. w Models that use a static Q10 or Rd parameters. In spite of the evidence that short-term temperature z Models that use a static Rd : A. responses in plants often do not fit the simple static Q10 PnET, photosynthesis and evapo-transpiration. exponential relationship, and although many modelers recognize the imperfections of the Q10 relationship, it is relationship between the temperatures 20 and 40 1C. still in wide use because until now there has not been a The TEM approach, however, does not capture the clear and generalizable alternative. All but three of the whole response of the general relationship described by 19 published models that we reviewed used either a Tjoelker et al. (2001). static Q10, parameter, a static Rd parameter, or both The purpose of this study was to test the degree to (Table 1). Moreover, despite evidence of thermal which variation in short-term temperature–respiration acclimation of Rdref, none of these models include functions and acclimation to temperature in these acclimation. In attempts to circumvent some of these functions influence total foliar respiratory flux from a problems, models have been developed that do not broad range of terrestrial ecosystems. To do so, we calculate respiration at all, but assume net primary simulated C balances for a range of vegetation types production (NPP) is a fixed proportion of gross primary and across a climate gradient using photosynthesis and production (GPP) (Coops et al., 1998; Waring et al., 1998; evapo-transpiration (PnET), a physiologically based, Waring & McDowell, 2002). This approach has been process-oriented ecosystem model (Aber et al., 1995, used to estimate leaf area and GPP at the stand scale 1996). We selected four ecosystems representing diverse (Waring & McDowell, 2002), estimate volume growth at climate and vegetation types. We examined the con- the landscape scale (Coops & Waring, 2001), estimate sequences of alternative temperature response and global GPP (Gifford, 2003), and suggests that warming respiration acclimation algorithms on modeled C of the biosphere has had little effect, thus far on budgets under both historic climate and climate autotrophic C emissions (Gifford, 2003). However, the warming scenarios. Given the inherent logic in most approach may fail to account for site-level climatic and of the reviewed models, our results should be qualita- plant functional variation. In addition, C balance details tively general and applicable. that are important at local scales and exist with the range of observed R : P ratios may be lost in global Methods averages. Another alternative to a fixed Q10 has been to describe more than one Q10–temperature relationship. We compared simple static respiration parameters and For example, the TEM model (Raich et al., 1991) uses alternative respiration algorithms within model simu- one relationship between 0 and 5 1C, the constant 2.0 lations across a range of vegetation types and sites. We between the temperatures 5 and 20 1C, and a second chose sites that represented a 201 latitudinal range, with r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
438 K . R . W Y T H E R S et al. concomitant changes in vegetation and climate. We Table 2 Site-specific PnET parameters for a black spruce site modified PnET parameters (from Aber et al., 1995) for: at Northern BOREAS (NOBS), an eastern hardwoods site at (1) Northern Boreas (NOBS), a coniferous boreal forest Harvard Forest (HARV), a tall-grass prairie site Konza Prairie (551N, 981W, near Thompson, MB, Canada) dominated (KONZ), and a broad leaved deciduous site at Coweeta by a Picea mariana overstory with some occasional Larix Hydrologic Laboratory (COWET) laricina, and a sparse occurrence of Pinus banksiana and Parameter NOBS HARV KONZ COWET Populus balsamifera, (2) Harvard Forest (HARV), a cold temperate hardwood forest (421N, 721W, near Peter- Site variable sham, MA, USA) dominated by Acer saccharum mixed Lat (1) 55 42 39 35 with Quercus rubra, Fraxinus spp., Tilia spp., and Fagus Whc 12 12 20 18 spp., and an understory comprised of saplings of Canopy variables K 0.5 0.58 0.68 0.58 shade-tolerant species and Vaccinium spp., (3) Konza folNCon (%) 0.8 1.9 1.6 2.0 Prairie (KONZ), a tall-grass prairie (391N, 961W, near SlwMax 200 100 100 100 Manhattan, KS, USA) dominated by a mixture of SlwDel 0 0.2 0.2 0.2 Andropogon gerardii, Schizachyrium scoparium, Sorghas- folReten (years) 6 1 1 1 trum nutans, and assorted forbs, and (4) Coweeta GddFolStart 300 100 550 600 Hydrologic Laboratory (COWET), a cool temperate GddFolEnd 1200 900 1800 900 hardwood forest (351N, 831W, near Otto, NC, USA) GddWoodStart 300 100 0 1000 representing a low elevation site comprised of Quercus GddWoodEnd 1200 900 0 1500 prinus, Acer rubrum, Liriodendron tulipifera, Carya spp., Photosynthesis variables and other Quercus spp. We used 5 years of local climate AmaxA 5.3 46 75 46 data from each site (1994–1998) (see Table 2 for AmaxB 21.5 71.9 190 71.9 BaseFolRespFracw 0.1 0.1 0.1 0.1 parameter values). Daytime and night-time interannual HalfSat 250 200 300 300 growing season temperatures were reasonably consis- AmaxFrac 0.76 0.75 0.8 0.75 tent for all 5 years at each site (Table 3). Across sites, PsnTOpt 20 24 28 24 daytime and night-time temperatures were lowest at PsnTMin 0 2 4 2 NOBS and increased for HARV, KONZ, and COWET, respQ10z 2 2 2 2 respectively. Water balance variables The canopy subroutines of PnET are constructed dvpd1 0.05 0.05 0.05 0.05 around a group of algorithms that apply physiological dvpd2 2 2 2 2 relationships between foliar nitrogen, photosynthetic PrecIntFrac 0.15 0.11 0.06 0.11 capacity, vertical scaling of leaf mass area, and leaf life- WueConst 10.9 10.9 46.5 10.9 span (Reich et al., 1992, 1994; Gower et al., 1993; FastFlowFrac 0.09 0.1 0.1 0.1 F 0.04 0.04 0.04 0.04 Ellsworth & Reich, 1993; Aber et al., 1995, 1996). Allocation variables Phenology is controlled by a cumulative heat sum cFracBiomass 0.45 0.45 0.45 0.45 algorithm (Aber et al., 1996). The model adds foliage RootAllocA 0 0 0 0 mass when growing degree day conditions are met. Leaf RootAllocB 2 2 2 2 off follows Aber et al. (1996), dropping leaves based on GRespFrac 0.25 0.25 0.25 0.25 each canopy layer’s C balance and a limit which RootMRespFrac 1 1 2 1 prevents senescence from occurring before a particular WoodMRespFrac 0.04 0.07 0.07 0.07 day. We ran the model for 100 years with randomized PlantCReserveFrac 0.75 0.75 0.75 0.75 and repeated climate data from local meteorological w Indicates a parameter for which Eqn. (1) was substituted in stations for each site in order to stabilize C pools before acclimation simulations, and zIndicates a parameter for which we output data used for these comparisons. Eqn. (2) was substituted in acclimation simulations. See Aber In its original form, PnET uses a Rd parameter fixed et al. (1995, 1996) for parameter definitions. at 10% of leaf net photosynthetic capacity (Amax) at 20 1C and a Q10 of respiration fixed at 2.0 (Aber et al., 1996). To test the effects of temperature acclimation on of basal respiration as a proportion of leaf Amax as dark respiration and the effects of a temperature- follows variable Q10, we substituted new algorithms for basal Rdacclim ¼ Amax ½0:14 0:002 T; ð2Þ respiration and Q10 of respiration individually and in combination. In order to simulate the effects of where Rdacclim is dark respiration, Amax is photosyn- temperature acclimation of respiration at the leaf level, thetic capacity, and T is temperature ( 1C). Conse- we introduce a simple linear temperature dependence quently, Rdacclim as a proportion of Amax declines with r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
F O L I A R R E S P I R AT I O N A C C L I M AT I O N A LT E R S E C O S Y S T E M C A R B O N B A L A N C E 439 Table 3 Mean daytime and night-time growing season temperatures ( 1C) for four sites (growing season as defined by days for which the mean temperature is greater than 0 1C) NOBS HARV KONZ COWET Year Night Day Night Day Night Day Night Day 1994 3.8 16.2 12.8 22.5 15.9 26.9 16.3 28.8 1995 3.5 18.0 11.5 19.9 15.8 27.8 15.5 26.7 1996 4.3 18.1 12.4 21.3 16.1 27.3 16.1 27.6 1997 4.2 18.4 13.2 20.8 16.0 27.6 15.3 27.7 1998 5.0 19.0 13.5 22.0 15.3 27.7 17.7 29.2 x 4.0 17.9 12.7 21.3 15.8 27.5 16.2 28.0 See Table 2 for site acronyms. increasing temperature. Assuming no appreciable accli- We modified PnET-II to operate on a daily time-step mation in leaf Amax to ambient temperature (or minimal in order to redefine the response each successive night in comparison with acclimation in respiration), acclima- using Q10var and Rdacclim. Source code is available on tion in leaf respiration rate alone would result in a request. There is evidence that the shift of the declining proportion of leaf respiration to Amax with acclimation curve can occur over a period as short as increasing environmental temperature for leaves mea- one to several days (Atkin et al., 2000a, b; Bolstad et al., sured at the same temperature. The relationship was 2003). Thus, the elevation of the curve (i.e. Rdacclim) and derived from field-based measures of needle dark the mean temperature ( 1C) each night is used to set respiration in numerous Pinus banksiana populations Q10var . Together, these drive the complete dark respira- at three sites (MI, USA, MN, USA, and Ontario, Canada) tion adjustment. across seasons (M. Tjoelker, J. Oleksyn, P. Reich, We output 5 years (1994–1998) of predicted daily unpublished data). In that dataset, specific respiration night-time respiration from foliage, annual total mass- rates at a standard temperature ranged from 20% and area-based foliar respiration, foliar respiration to higher to 20% lower than the seasonal mean photosynthesis ratios (Rd : A), and annual above ground ( 5 4.0 nmol g1 s1, s 5 0.6) across the range of mean NPP (ANPP). To develop the necessary comparisons, we daily air temperatures of 9–21 1C. Although there is ran the model with the original static respiration evidence that species differ in temperature acclimation parameters for each of the above scenarios, then reran to dark respiration, the magnitude (proportional change) the model after substituting the Rdacclim algorithm only, of temperature acclimation of respiration observed for then again with the Q10var algorithm only, and finally with jack pine and used in this study is generally comparable both new algorithms simultaneously. In order to with findings in studies of oak, pine, and other examine the effects of the Rdacclim and Q10var algorithms temperate and boreal species (Tjoelker et al., 1999a; on temperature warming scenarios, we ran the model in Atkin et al., 2000b; Bolstad et al., 2003). Few studies have both static-and modified respiration-parameter modes examined leaf level Rd : A ratios in relation to variation in with historic climate and under a simple 1 2 1C (day- growth or ambient temperatures, although it is thought time and night-time) temperature warming scenario. that Rd : A may vary within a limited range (Reich et al., 1998a, b; Cannell & Thornley, 2000). Our purpose here is Results to test the consequences of assumed temperature acclimation effects on leaf level Rd : A for ecosystem Temperature response scale carbon balance. The dependence of Q10var on Predicted respiration rates using alternative respiration measurement temperature is from Tjoelker et al. (2001): algorithms differed from respiration estimates based on Q10var ¼ 3:22 0:46T; ð3Þ static parameters and there was variation in magnitude and shape of the respiration temperature response where Q10var is the Q10 of foliar respiration and T is among sites (Fig. 1). The decreases in predicted temperature ( 1C). From here on, we will refer to these respiration, averaged across seasons and substituting new relationships as Rdacclim and Q10var algorithms, both alternative respiration algorithms, ranged from respectively. Collectively, the new algorithms make a 30% for a temperate hardwood forest at COWET to 60% net contribution of two additional parameters to the for a tall-grass prairie at KONZ (Tables 3 and 4). PnET model. Although temperature acclimation should conserve r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
440 K . R . W Y T H E R S et al. Fig. 1 Canopy dark respiration response to night-time temperature. Gray circles represent 5-year model runs of photosynthesis and evapo-transpiration (PnETs) static Rd and static Q10 parameters. Black x’s represent modeled temperature response data after substituting Rdacclim and Q10var individually and in combination. See Table 2 for site acronyms. foliar respiratory C during warm periods and release C ranged from less than 0.1 mg C g foliage1 day1 at during cool periods, the net effect was to conserve C NOBS, to 2.5 mg C g foliage1 day1 at KONZ. How- across the entire temperature trajectory. ever, while respiration estimates were consistently high Predicted respiration rates after substituting Rdacclim across sites at temperatures between 0 and 5 1C in alone were lower at high temperatures, and higher at simulations that used static parameters, the generally low temperatures, relative to estimates based on static low respiration rates at these temperatures suggest that parameters (Fig. 1). While NOBS showed small differ- any differences in rates of respiratory C loss on the ences in predicted respiration rates based on Rdacclim overall C balance might be a relatively small propor- relative to estimates based on static parameters, respira- tional change in terms of C on a per day basis. tion differences were larger under warmer conditions at Respiration estimates that used both the Rdacclim and the warmer sites. Predicted respiration rates calculated Q10var algorithms in combination were consistently and after substituting Rdacclim diverged from those calculated substantially lower across the entire annual tempera- with static parameters above 18 1C and below 5 1C. At ture range in simulations, relative to model runs using temperatures greater than 18 1C, respiration rates esti- static parameters (Fig. 1). Between 5 and 20 1C, the mated with the Rdacclim algorithm resulted in differences respiration estimates based on the alternative respira- that ranged from 0.5 mg C g foliage1 day1 lower at tion algorithms were roughly half as large as the static NOBS, to 4.0 mg C g foliage1 day1 lower at KONZ. parameter estimates at all sites. Below 5 1C, respiration differences estimated with Rdacclim were slightly higher at all sites. Foliage mass At all sites, respiration after substituting the Q10var algorithm alone resulted in markedly reduced Foliage mass predictions were similar for model runs respiration rates at the low end of the temperature using the alternative Rdacclim and Q10var algorithms and response curve compared with those that used the for those using static parameters for three of the four static Q10 parameters (Fig. 1). Both estimates converged sites (Fig. 2), implying that alternative respiration at the upper end of the temperature range. Differences algorithms had modest effect on canopy size. Foliage in dark respiration (because of altered Q10) near 0 1C mass estimates were 8% higher at COWET using the r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
F O L I A R R E S P I R AT I O N A C C L I M AT I O N A LT E R S E C O S Y S T E M C A R B O N B A L A N C E 441 Fig. 2 Predicted foliage mass with and without alternative respiration algorithms for all sites using measured climate and temperature warming scenarios ( 1 2 1C). Error bars represent 1 standard deviation. See Table 2 for site acronyms. Table 4 Annual foliar respiration on a leaf mass basis (g C g leaf1 yr1) predicted by PnET with static parameters (Static) and alternative (Altrn) respiration response algorithms NOBS HARV KONZ COWET Ambient 12 Ambient 12 Ambient 12 Ambient 12 Year Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn 1994 0.19 0.11 0.22 0.13 1.25 0.75 1.46 0.89 3.95 2.47 4.67 2.86 1.81 1.24 2.17 1.44 1995 0.19 0.11 0.21 0.13 1.29 0.76 1.53 0.89 3.94 2.29 4.65 2.60 2.02 1.23 2.16 1.40 1996 0.19 0.11 0.22 0.13 1.22 0.75 1.44 0.86 3.78 2.25 4.47 2.66 1.89 1.16 2.41 1.41 1997 0.20 0.12 0.23 0.14 0.92 0.55 1.09 0.66 3.74 2.21 4.45 2.59 1.94 1.19 2.40 1.41 1998 0.21 0.12 0.24 0.14 1.05 0.67 1.26 0.80 4.32 2.51 5.15 2.98 2.16 1.40 2.30 1.60 x (V) 0.20(5) 0.11(5) 0.22(4) 0.13(4) 1.14(4) 0.70(13) 1.36(13) 0.82(12) 3.95(6) 2.35(6) 4.68(6) 2.74(6) 1.96(7) 1.24(7) 2.29(5) 1.45(6) Measured climate is represented by Ambient. Two degree centigrade increase is represented by 1 2. Means are represented by x , coefficients of variation (as percent) are in parentheses. See Table 2 for site acronyms. PnET, photosynthesis and evapo-transpiration. alternative algorithms relative to the static respiration (Table 4). Results were consistent on both a mass basis parameters. Foliage mass (per unit ground area) and on a ground area basis (Table 5). decreased by approximately 10% for NOBS, HARV, Modification of model respiration algorithms had a and KONZ in elevated temperature scenario simula- larger impact on estimated Rd than simulated warming. tions, but there was no change in foliage mass at Five year mean respiration estimates, under a 1 2 1C COWET with elevated temperature. warming scenario, were roughly 5–15% higher for all sites when using either the static parameter model for Annual foliar respiration both climate scenarios, or the alternative respiration Alternative respiration algorithms reduced annual algorithms for both scenarios. However, the respiration respiration compared with estimates based on static estimates using alternative respiration algorithms (i.e. parameters. Annual foliar respiration estimates using Rdacclim plus Q10var ) within the temperature warming both Rdacclim and Q10var algorithms in combination were ( 1 2 1C) scenarios were substantially lower than the lower in all years, at all sites, and for both ambient static parameter estimates from the ambient tempera- temperature and temperature warming scenarios when ture simulations. These decreases suggest that model compared with simulations using static parameters sensitivity to physiological algorithms is on a similar r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
442 K . R . W Y T H E R S et al. Table 5 Annual foliar respiration on an area ground basis (g C m2 ground yr1) predicted by PnET with static parameters (Static) and alternative (Altrn) respiration response algorithms NOBS HARV KONZ COWET Ambient 12 Ambient 12 Ambient 12 Ambient 12 Year Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn 1994 152 91 158 98 299 197 342 223 1241 859 1383 892 480 325 460 330 1995 148 87 156 94 299 199 347 220 1146 754 1240 773 449 322 531 352 1996 150 90 156 95 289 194 337 216 1143 748 1242 767 462 313 528 362 1997 158 94 163 99 191 140 236 165 1290 817 1405 823 458 319 556 372 1998 167 99 175 105 251 177 296 204 1368 857 1480 882 507 350 574 388 x (V) 155(5) 92(5) 162(5) 98(4) 266(17) 181(14) 312(15) 206(12) 1237(8) 807(7) 1350(8) 827(7) 471(5) 326(4) 530(8) 361(6) Measured climate is represented by Ambient. Two degree centigrade increase is represented by 1 2. Means are represented by x , coefficients of variation (as percent) are in parentheses. See Table 2 for site acronyms. PnET, photosynthesis and evapo-transpiration. Fig. 3 Change in annual area based respiration illustrating (a) the decrease in respiration for ambient and elevated temperature simulations as a result of alternative respiration algorithms and (b) the increase in respiration for static parameter and alternative algorithm simulations as a result of elevated temperature. See Table 2 for site acronyms. order or larger than effects of substantial climate Rd : A ratio warming. The effects of alternative respiration algorithms on Rd : A ratios were consistently smaller ( 40% less) in annual respiration tended to be relatively greater at the simulations that used Rdacclim and Q10var algorithms warmer sites, especially at KONZ (Fig. 3). The effects of relative to estimates from static parameters (Fig. 4). In alternative respiration algorithms were slightly greater addition, Rd : A ratios tended to be higher at sites with with climate warming relative to the ambient tempera- higher mean night- and daytime temperatures. Rd : A ture simulations at three of the four sites. The effects of ratios from simulations that used Rdacclim and increased temperature on annual respiration tended to Q10var algorithms tended to be less than Rd : A ratios be greater at the warmer sites, and highest at KONZ, in from simulations that used static respiration para- simulations that used static parameters. However, the meters. Predicted Rd : A from simulations that used temperature enhancement in warming simulations that Rdacclim and Q10var r algorithms ranged from 0.15 at used alternative respiration algorithms was notably NOBS to 0.27 at COWET. Predicted Rd : A from static lower in three of the four sites. . These results suggest respiration parameters ranged from 0.25 at NOBS to that incorporating realistic models of respiration is 0.43 at COWET. Climate warming appeared to have no important, since response to climate warming, appears effect on Rd : A estimates from simulations that used to vary with model characteristics. alternative respiration algorithms, while climate warm- r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
F O L I A R R E S P I R AT I O N A C C L I M AT I O N A LT E R S E C O S Y S T E M C A R B O N B A L A N C E 443 Fig. 4 Rd : A ratios for 5 years, with and without alternative respiration algorithms, using measured climate and temperature warming scenarios ( 1 2 1C). Error bars represent 1 standard deviation among years. See Table 2 for site acronyms. Table 6 ANPP (g biomass m2 yr1) predicted by PnET with static parameters (Static) and alternative (Altrn) respiration response algorithms NOBSw HARV KONZ COWET Ambient 12 Ambient 12 Ambient 12 Ambient 12 Year Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn Static Altrn 1994 358 478 394 512 677 843 722 945 431 526 383 464 704 1008 587 999 1995 348 468 390 508 679 852 726 956 416 502 361 453 660 999 636 1022 1996 363 484 388 507 673 851 697 941 401 486 348 451 677 995 620 1007 1997 340 459 370 487 624 802 645 891 409 497 354 439 680 994 652 1039 1998 344 464 384 503 640 789 682 889 430 530 346 450 746 1054 692 1099 x (V) 351(3) 471(2) 385(2) 503(2) 659(4) 827(4) 694(5) 924(3) 417(3) 508(4) 358(4) 451(2) 693(5) 1010(2) 637(6) 1033(4) w Does not include bryophyte contribution. Means are represented by x, coefficients of variation (as percent) are in parentheses. See Table 2 for site acronyms. See Table 4 for descriptions of Ambient and 1 2. PnET, photosynthesis and evapo-transpiration; ANPP, above ground net primary production. ing introduced larger Rd : A estimates in simulations forested sites. Moreover, it is notable that the two model using static respiration parameters. types resulted in variable responses to climate warming at some sites. For example, annual ANPP at NOBS ANPP increased similarly (33 g biomass m2 yr1) with warm- ing in both fixed parameter simulations and in the Predicted ANPP from alternative respiration algo- alternative algorithm simulations (Rdacclim and Q10var ), rithms was higher in all years, under both climate while annual ANPP at KONZ decreased in response to scenarios, and at all sites relative to predictions from warming by approximately the same amount static parameters (Table 6). Response of ANPP to (58 g biomass m2 yr1) in both the static parameter alternative respiration algorithms and to temperature and the alternative algorithm simulations. At the two were variable among ecosystems (Figure 5). ANPP hardwood sites (HARV and COWET), response of response to alternative respiration algorithms (with simulated annual ANPP to climate warming was more both Rdacclim and Q10var ) were larger at warmer sites positive using Rdacclim and Q10var algorithms compared than cooler sites, but smaller at the prairie site than all with simulations that used the static respiration r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
444 K . R . W Y T H E R S et al. Fig. 5 Changes in above ground net primary production (ANPP) at (a) two different temperatures under the effects of alternative respiration algorithms, and (b) for alternative respiration algorithms and static parameters under the effects of elevated temperature. See Table 2 for site acronyms. parameters. Annual ANPP at HARV increased with model, despite similar basal rates. Conversely, at high warming by 35 g biomass m2 yr1 in the original temperatures in the combined model, lower basal rates version, but increased by 97 g biomass m2 yr1 in the because of acclimation result in lower respiration rates modified version. Annual ANPP at COWET decreased than the static model, despite similar Q10’s. However, by 56 g biomass m2 yr1 with warming in the static the magnitude of these effects appear to be dependent respiration parameter version, but increased upon specific site environmental differences and may 23 g biomass m2 yr1 with warming in simulations that also reflect variation in concomitant plant physiology. used the modified algorithms. The changes in respiration because of model modifaca- tion seen in Fig. 1 lead to associated changes in annual total respiration estimates. Total respiration on both a mass Discussion basis and a ground area basis were lower in simulations using the Rdacclim and Q10var algorithms compared with Alternative algorithms: Rdacclim and Q10var simulations using the static parameters. Decreases in While substituting alternative respiration algorithms for annual respiration on a ground area basis that resulted static parameters reduced simulated plant respiration from substituting Rdacclim and Q10var algorithms were 41%, across a range of temperatures at all sites, this effect 36%, 35%, and 31% for NOBS, HARV, KONZ, and varied among sites (as seen in Fig. 1). The consequences COWET, respectively. These large differences in estimated of Rdacclim and Q10var together on modeled carbon budgets carbon efflux between the original and alternative models appears to be a consistent and substantial reduction in support the conclusions of Gunderson et al. (2000) and estimated respiratory carbon losses across the entire Tjoelker et al. (2001) that ecosystem models should temperature range. Predicted respiration rates at very incorporate basal respiration acclimation and tempera- low temperatures should be low despite a high Q10 ture-variable Q10 relationships, particularly if the model is observed at low temperatures (Tjoelker et al., 2001). This to be applied across a large spatial extent where broad is because most models use respiration rates measured ranges in climate are to be expected. at relatively high temperatures (such as 20 1C) as their While it should be noted that anytime model starting point. In order to have a higher Q10 at lower complexity is increased, an associated increase in temperatures and converge on the same respiration rate uncertainty is risked. In this case we have increased at 20 1C, rates must be lower at very low temperatures the number of parameters in the PnET model (one than would be otherwise expected. additional parameter in the basal respiration calcula- It is notable that the combined effects of the Rdacclim tion, and one additional parameter in the Q10 calcula- and Q10var algorithms appear greater than expected from tion). In justifying this change, we would argue that a summing of their individual effects. This interaction is removing the parameters BFolResp and RespQ10 and likely because at low temperatures in the combined replacing them with simple, generalizable, and biolo- model, the Q10var algorithm causes large differences in gically realistic algorithms that appear to hold up predicted respiration rates compared with the static across biomes and across a broad range of taxa, is a r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
F O L I A R R E S P I R AT I O N A C C L I M AT I O N A LT E R S E C O S Y S T E M C A R B O N B A L A N C E 445 Table 7 Measured ANPP means from literature; using longest available records for sites as similar to and as close as possible to stands used in the modeling Site ANPP (g m2) Length of record Source NOBS 487* 3 Bond-Lamberty et al. (2001) HARV 745 8 Knapp & Smith (2001) KONZ 528w 21 Knapp et al. (1998) COWET 1110z 10 Bolstad et al. (2001) *Well drained soils, stand age 420 years, does not include bryophyte contribution. w Annually burned lowlands. z Calculated for 685 m elevation. See Table 2 for site acronyms. ANPP, above ground net primary production. complexity vs. generalizability trade-off that is worth and of equally good fit at the other. At NOBS, KONZ, making. In addition, we would argue that ecosystem and COWET, estimated ANPP using the modified models should be rooted in simple biological mechan- respiration algorithms were within 3%, 4%, and 9% of isms wherever possible, and the new algorithms have a field observations, while estimates using the static much stronger physiological basis than the earlier ones. respiration algorithms were within 28%, 21%, and Finally, the reductions in foliar respiration that 38% of field observations, respectively. At the fourth resulted from our new algorithms were of a magnitude site, HARV, the two estimates were equally close (both that would explain the overestimated foliar respiration were within about 11% of field observations). However, value relative to measured field data reported by Law there is sufficient variability (and error) in measured et al. (2000) for PnET-II using static respiration para- ANPP across local gradients, among years, and among meters in a Pinus ponderosa system. Additionally, the reports, that it is problematic to use reported field nature of Rd acclimation and its importance as shown ANPP to assess the reliability of the static or alternative in models here in are consistent with work suggesting model output. Furthermore, given the fact that a broad Rd acclimation across global climate gradients simulation model can yield good agreement with (Enquist et al., 2003). measured field data through a collection of compensat- Variation in reduced C loss to respiration on an annual ing errors, one must be careful to not take model basis among sites appears to be not only a function of the agreement alone as the singular reason to modify (or to relative shape of the temperature response curve, but not modify) these kinds of process based ecosystem also the amount of time over the course of the season models. Given that the alternative respiration algo- spent at any given end of the curve. In other words, the rithms presented here incorporate well documented respiration–temperature response curve represents all biological processes, which appear generalizable across thermal environments over the entire year. The amount biomes and a broad range of taxa, and appear to fit with of total C conserved on an annual basis will depend published empirical data as well as or better than the upon the relative proportion of the year spent at cold vs. static parameter output, these algorithms appear useful warm ends of the temperature curve, and the relative and should be pursued further. shapes of the curves. For example, at sites with short The higher ANPP estimates for historic climate runs growing seasons, such as NOBS, a large portion of the of 25%, 20%, 18%, and 31% at NOBS, HARV, KONZ, year will be spent at temperatures well below the and COWET, respectively, from simulations using the physiological optimum for the species present, whereas Rdacclim and Q10var algorithms result from the represen- sites with a longer growing season, such as COWET, a tation of C conserved through respiration acclimation greater portion of the year will be spent at temperatures to temperature (Table 6). At the three forested sites, that favor physiologic activity. reduced respiratory efflux is almost directly translated Although field measurements of ANPP can include into increased ANPP. However, at the prairie site, there substantial uncertainties, they may be useful to com- is a greater amount of C conserved via the Rdacclim and pare with ANPP estimates from the static respiration Q10var algorithms than is represented in increased ANPP vs. alternative respiration versions of PnET. Estimates (430 g C m2 yr1 vs. 91 g biomass m2 yr1, given a of ANPP calculated from the combined alternative parameter of 0.45 to convert C to biomass, see Table respiration algorithms were a closer match to published 2). In the PnET model, the difference between con- ANPP field data than ANPP based on static respiration served respiratory C and the additional C allocated to parameters at three of the four sites (see Tables 6 and 7) ANPP accumulates in the general plant C pool. There- r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
446 K . R . W Y T H E R S et al. fore, once C is conserved through respiratory path- substantial implications for the majority of commonly ways, variation in allocation strategies among plant used ecosystem models. types (species) could affect how and where that C is partitioned. In addition, while Tjoelker et al. (2001) Climate change developed Q10var from data that included grasses and Given the importance of predicted climate change, the forbs, Rdacclim was based on data from trees only. For contrasts in respiration response to warming between this reason Rdacclim may not well represent dark the two model versions (alternative vs. static para- respiration acclimation to temperature in the vegetation meters) are of particular interest. Relative to historic types prominent at KONZ. climate, elevated temperature simulations increased Although Rd : A ratios appear to scale to some degree, annual respiration by between 1% and 12% at all four they are variable at the leaf level (Reich et al., 1998a, b; sites using both versions of the model, but the influence Amthor, 2000), and at the whole plant level (Gifford, of model modification on these varied among sites. 2003), and published empirical data are not yet sufficient Predicted foliar respiration with static respiration to comprehensively determine the degree to which Rd : A parameters increased by 8% and 11% under the climate ratios are constrained across species and environments. warming scenario at KONZ and COWET, respectively, Rd : A is of interest because Rd : A ratios help place but when modified respiration algorithms were incor- respiration in the context of an C balance, and aid in porated into the model, warming increased predicted interpreting ANPP. Because implementing Rdacclim and foliar respiration by only 2% and 1% at those same sites. Q10var algorithms in PnET-reduced respiration, and PnET In contrast, at NOBS and HARV (the cooler pair of our does not include direct acclimation of photosynthesis to four sites) the predicted increases in respiration because temperature, lower Rd : A ratios are expected. Here an of warming, were similar using both versions of the important question is whether photosynthesis acclimates model. This suggests that models that incorporate or adjusts to increasing temperature and whether this is respiration acclimation and temperature variable Q10 similar to Rd. However, given the distinctly different algorithms may yield different predictions of respira- shapes of their temperature response curves, in the short tion response to climate warming than models using term photosynthesis is relatively constant across a broad static respiration parameters, and that such differences range of temperatures relative to Rd (Chabot and Lewis, may vary across systems. These particular observations 1976, Aubuchon et al., 1978, Jurik, 1986, Jurik et al., 1988; could be in part because of down-regulation of Gunderson et al., 2000). Moreover, thermal acclimation respiration rates at higher temperatures, which are for photosynthesis exhibits variable patterns among more common at the two warmer sites. This connection species (Slatyer & Morrow, 1977; Dougherty et al., 1979; between thermal environment and respiration suggests Jurik et al., 1988; Ferrar et al., 1989) and has not been well that down-regulation of respiration could have a correlated with climate (Tranquillini et al., 1986; Ferrar et substantial impact on C balance and productivity, al., 1989). It is thus, still unclear how to best represent particularly at warm sites. The decrease in ANPP from photosynthetic acclimation in an ecosystem model such the effects of elevated temperature at KONZ (Fig. 5b) as PnET. may be in part because of KONZ being a more water- Although respiration in plants is known to acclimate limited system than the three forest sites, or because of (hours to days) to temperature (Tjoelker et al., 1999a, b; physiologic differences in vegetation, or because of Atkin et al., 2000b, Bolstad et al., 2003), and a general PnETs allocation logic, or any combination of the above. linear relationship describing the short-term tempera- Our predictions of increasing Rd : A ratios from ture dependence of Q10 for foliar respiration appears to colder to warmer climates, and from standard climate hold across biomes (Tjoelker et al., 2001), the data on simulations to 1 2 simulations, support the experi- which Rdacclim and Q10var algorithms were based origi- mental findings of Tjoelker et al. (1999a) that Rd : A nated from broad thermal gradients. Tjoelker et al. tended to increase with warming in boreal-tree seed- (2001) reported mean Q10 values that ranged from 2.14 lings. This may be because differences in the degree to to 2.56 (tropical to arctic biomes, respectively). Tjoelker which plant growth and size, growth respiration, and et al. (2001) suggested the need for more common maintenance respiration respond to warming could temperature studies to isolate the effects of measure- affect Rd : A ratios (Amthor, 2000), or be complicated by ment temperature from acclimation. In addition, others the effects of water stress (Cannell & Thornley, 2000). have suggested that while some plant species exhibit a high degree of acclimation, some do not (e.g. Lar- Conclusions igauderie & Körner, 1995). Nonetheless, our findings depart substantially from modeled ecosystem fluxes Because the instantaneous temperature response func- based on static respiration parameters and have tion of plants is temperature dependent (Tjoelker et al., r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
F O L I A R R E S P I R AT I O N A C C L I M AT I O N A LT E R S E C O S Y S T E M C A R B O N B A L A N C E 447 2001) and thermal acclimation of respiration appears to and grassland ecosystems. Ecological Applications, 1, 118– be common (Bolstad et al., 2003; Tjoelker et al., 1999a, b; 138. Atkin et al., 2000a,b; Luo et al., 2001), our results suggest Amthor JS (2000) The McCree–de Wit–Penning de Vries– that: (1) Rdacclim and Q10var algorithms allow ecosystem Thornley respiration paradigms: 30 years later. Annals of Botany, 86, 1–20. models to account for respiration response to tempera- Atkin OK, Edwards EJ, Loveys BR (2000a) Response of root ture in a more biologically realistic way than static respiration to changes in temperature and its relevance to parameter models can, (2) simple, generalized Rdacclim global warming. New Phytologist, 147, 141–154. and Q10var algorithms alter modeled estimates of C Atkin OK, Holly C, Ball MC (2000b) Acclimation of snow gum exchange and production in plants and ecosystems (Eucalyptus pauciflora) leaf respiration to seasonal and diurnal and alter plant responses to warming scenarios, (3) variations in temperature: the importance of changes in the incorporating Rdacclim and Q10var into process-based eco- capacity and temperature sensitivity of respiration. Plant Cell system models is important in that the effects of Rdacclim and Environment, 23, 15–26. and Q10var appear to be as large or larger than the effects Atkin OK, Tjoelker MG (2003) Thermal acclimation and the of major climate change and appears to be consistent dynamic response of plant respiration to temperature. Trends with measured ANPP and eddy covariance estimates in Plant Science, 8, 343–351. (Table 7, Law et al 2000, Enquist et al. 2003), (4) current Aubuchon RR, Thompson DR, Hinckley TM (1978) Environ- mental influences on photosynthesis within the crown of a models may over-predict respiration and result in either while oak [Quercus alba]. Oecologia, 35, 295–306. underestimated or inappropriately allocated productiv- Bartelink HH (2000) Effects of stand composition and thinning ity estimates, especially on warm sites or when running in mixed-species forests: a modeling approach applied to climate warming scenarios, and (5) algorithms that Douglas-fir and beach. Tree Physiology, 20, 399–406. incorporate plant acclimation to temperature may Belehrádek J (1930) Temperature coefficients in biology. Biologi- enhance the use of ecosystem models across broad cal Reviews, 5, 30–58. spatial scales or with simulated temperature change Berry JA, Raison JK (1981) Response of macrophytes to where large temperature ranges may cause erroneous temperature. In: Encyclopedia of Plant Physiology, New Series, results from models that use static respiration para- Vol. 12A (eds Lange OL), pp. 277–338. Springer, Berlin. meters. Finally, while these results represent incorporat- Bolstad PV, Mitchell K, Vose JM (1999) Foliar temperature- ing Rdacclim and Q10var algorithms into the PnET model, it respiration response functions for broad-leaved tree species in the southern Appalachians. Tree Physiology, 19, 871–878. is likely that many other physiologically based ecosys- Bolstad PV, Vose JM, McNulty SG (2001) Forest productivity, leaf tem models that utilize static respiration parameters area and terrain in southern Appalachian deciduous forests. would yield similar results if similarly adjusted. Forest Science, 47, 1–9. Bolstad PV, Reich PB, Lee T (2003) Rapid temperature acclima- Acknowledgments tion of leaf respiration rates of Quercus alba (L.) and Quercus rubra (L.) and Q. rubra. Tree Physiology, 23, 969–976. This research was supported by the NASA Terrestrial Ecology Bond-Lamberty B, Wang C, Gower ST (2001) Net primary Program (NAG 5-11222), NSF (IBN-9806628), NSF LTER pro- production and net ecosystem production in boreal black spruce gram (DEB-0080382), and the Wilderness Research Foundation. wildfire chronosequence. Global Change Biology, 10, 473–487. Cannell MGR, Thornley JHM (2000) Modelling the components References of plant respiration: some guiding principles. Annals of Botany, Aber JD, Federer CA (1992) A generalized, lumped parameter 85, 45–54. model of photosynthesis, evapotranspiration, and net primary Chabot BF, Lewis AR (1976) Thermal acclimation of photosynth- production in temperate and boreal forest ecosystems. esis in northern red oak [Quercus rubra borealis]. Photosynthe- Oecologia, 92, 463–474. tica, 10, 130–135. Aber JD, Ollinger SV, Driscoll CT (1997) Modeling nitrogen Clark KL, Cropper WP, Gholz HL (2001) Evaluation of modeled saturation in forest ecosystems in response to land use and carbon fluxes for a slash pine ecosystem: SPM2 simulations atmospheric deposition. Ecological Modelling, 101, 61–78. compared to eddy flux measurements. Forest Science, 47, 52– Aber JD, Ollinger SV, Federer CA et al. (1995) Predicting 59. the effects of climate change on water yield and forest Coops NC, Waring RH, Landsberg JJ (1998) Assessing forest production in the northern United States. Climate Research, 5, productivity in Australia and New Zealand using physiolo- 207–222. gically-based model driven with averaged monthly weather Aber JD, Reich PB, Goulden ML (1996) Extrapolating leaf CO2 data and satellite-derived estimates of canopy photosynthetic exchange to the canopy: a generalized model of forest capacity. Forest Ecology and Management, 104, 113–127. photosynthesis validated by eddy correlation. Oecologia, 106, Coops NC, Waring RH (2001) Estimating forest productivity in 267–275. the eastern Siskiyou Mountains of southwestern Oregon using Ågren GI, McMurtrie RE, Parton WJ et al. (1991) State-of-the-art a satellite driven process model, 3-PGS. Canadian Journal of models of production-decomposition linkages in conifer Forest Research, 31, 143–154. r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
448 K . R . W Y T H E R S et al. Cramer W, Kicklighter DW, Bondeau A et al. (1999) Comparing James WO (1953) Plant Respiration. Clarendon Press, Oxford. 282. global models of terrestrial net primary productivity (NPP): Jurik TW (1986) Seasonal patterns of leaf photosynthetic capacity overview and key results. Global Change Biology, 5, 1–15 (s1). in successional northern hardwood tree species. American Dougherty PM, Teskey RO, Phelps JE et al. (1979) Net Journal of Botany, 73, 131–138. photosynthesis and early growth trends of a dominate White Jurik TW, Briggs GM, Gates DM (1988) Springtime recovery of Oak (Quercus alba L.). Plant Physiology, 64, 930–935. photosynthetic activity of white pine in Michigan. Canadian Ellsworth DS, Reich PB (1993) Canopy structure and vertical Journal of Botany, 66, 138–141. patterns of photosynthesis and related leaf traits in a Keeling RF, Piper SC, Heimann M (1996) Global and hemi- deciduous forest. Oecologia, 96, 169–178. spheric CO sinks deduced from changes in atmospheric O2 Enquist BJ, Economo EP, Huxman TE et al. (2003) Scaling concentrations. Nature, 381, 218–221. metabolism from organisms to ecosystems. Nature, 423, 639– Këllomaki , Väisänen H (1997) Modelling the dynamics of the 642. forest ecosystem for climate change studies in the boreal Falkowski P, Scholes RJ, Boyle E et al. (2000) The Global Carbon conditions. Ecological Modelling, 97, 121–140. Cycle: a test of our knowledge of earth as a system. Science, Kimball JS, Keyser AR, Running SW et al. (2000) Regional 290, 291–296. assessment of boreal forest productivity using an ecological Farrar JF (1985) The respiratory sources of C02. Plant Cell and process model and remote sensing parameter maps. Tree Environment, 8, 427–438. Physiology, 20, 761–775. Ferrar JF, Slatyer RO, Vranjic JA (1989) Photosynthetic tempera- Knapp AK, Briggs JM, Blair JM et al. (1998) Patterns and controls ture acclimation in Eucalyptus species from diverse habitat, of above ground net primary production in tallgrass prairie. and a comparison with Nerium oleander. Australian Journal In: Grassland Dynamics: Long-Term Ecological Research in of Plant Physiology, 16, 199–217. Tallgrass Prairie (eds Knapp AK, Briggs JM, Hartnett DC, Field CB, Chapin FS, Matson PA et al. (1992) Responses of Collins SL), pp. 193–221. Oxford University Press, Oxford. terrestrial ecosystems to the changing atmosphere: a resource- Knapp AK, Smith MD (2001) Variation among biomes in based approach. Annual Review of Ecology and Systematics, temporal dynamics of above ground primary production. 23, 201–235. Science, 291, 481–484. Forward DF (1960) Effect of temperature on respiration. In: Landsberg JJ, Waring RH (1997) A generalized model of forest Handuch der Pflanzenphysiologie, Vol 12 (ed. Ruhland W), Part productivity using simplified concepts of radiation-use- 2, pp. 234–258. Springer, Berlin. efficiency, carbon balance, and partitioning. Forest Ecology Frolking S, Goulden ML, Wofsy SC et al. (1996) Modelling and Management, 95, 209–228. temporal variability in the carbon balance of a spruce/moss Larcher W (2003) Physiological Plant Ecology: Ecophysiology and boreal forest. Global Change Biology, 2, 343–366. Stress Physiology of Functional Groups (pp 120–126. Springer– Gifford RM (2003) Plant respiration in productivity models: Verlag, Berlin. conceptualization, representation and issues for global terres- Larigauderie A, Körner C (1995) Acclimation of leaf dark trial carbon-cycle research. Functional Plant Biology, 30, respiration to temperature in alpine and lowland plant 171–186. species. Annals of Botany, 76, 245–252. Gower ST, Reich PB, Son Y (1993) Canopy dynamics and Lavigne MB, Ryan MG (1997) Growth and maintenance aboveground production of five tree species with different leaf respiration rates of aspen, black spruce and jack pine stems longevities. Tree Physiology, 12, 327–345. at northern and southern BOREAS sites. Tree Physiology, 17, Gracia CA, Tello E, Sabate S et al. (1999) GOTWILA: an 543–551. integrated model of water dynamics and forest growth. In: Law BE, Waring RH, Anthoni PM et al. (2000) Measurements Ecology of Mediterranean Evergreen Oak Forests (eds Rodé F, of gross and net ecosystem productivity and water vapor Retana J, Gracia CA, Bellot J), Springer-Verlag, Berlin. exchange of a Pinus ponderosa ecosystem, and an evaluation of Gunderson CA, Norby RJ, Wullschleger SD (2000) Acclimation two generalized models. Global Change Biology, 6, 155–165. of photosynthesis and respiration to simulated climatic warm- Liu J, Chen JM, Cihlar J et al. (1997) A process-based boreal ing in northern and southern populations of Acer saccharum: ecosystem productivity simulator using remote sensing laboratory and field evidence. Tree Physiology, 20, 87–96. inputs. Remote Sensing of Environment, 62, 158–175. Gunn S, Farrar JF (1999) Effects of a 4 1C increase in temperature Lloyd J, Taylor JA (1994) On temperature dependence of soil on partitioning of leaf area and dry mass, root respiration and respiration. Functional Ecology, 8, 315–323. carbohydrates. Functional Ecology, 13, 12–20. Luan J, Muetzelfeldt RI, Grace J (1996) Hierarchical approach to Hari P, Mäkelä A, Berninger F et al. (1999) Field evidence for the forest ecosystem simulation. Ecological Modeling, 86, 37–50. optimality hypothesis of gas exchange in plants. Australian Lubchenco J A, Olson AM, Brubaker LB et al. (1991) The Journal of Plant Physiology, 26, 239–244. sustainable biosphere initiative: An ecological research agen- Houghton JT, Callander BA, Varney SK (1992) Climate Change: da. Ecology, 72, 371–412. The Supplementary Report to The IPCC Scientific Assessment. Luo Y, Wan S, Hui D et al. (2001) Acclimatization of soil Cambridge University Press, Cambridge. 200. respiration to warming in a tall grass prairie. Nature, 413, Houghton JT, Ding Y, Griggs DJ et al. (2001) Climate Change 2001: 624–625. The Scientific Basis. Cambridge University Press, Cambridge. Mäkelä A, Landsberg J, Ek AR et al. (2000) Process-based models 881. for forest ecosystem management: current state of the art and r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 435–449
You can also read